[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Changing CFLAGS for i386 packages on x86_64 (benchmarks included)



Dave Jones wrote:
On Sun, Jul 16, 2006 at 08:15:46AM -0400, Dave Jones wrote:
 > On Sun, Jul 16, 2006 at 02:10:22PM +0200, dragoran wrote:
> > I did same benchmarks using flops.c (attached) the results are attached > > to it seems that
 >  > -march=k8 -mtune=k8 -m32 is the fastes but not by much ...

Additionally, whilst you found these flags speed things up
slightly on AMD systems, it may have a noticable detrimental
impact on Intel systems, and shipping multiple sets of binaries
for a negligable speedup doesn't make a lot of sense.

this might be true but unfortunatelly I don't have a system to test (only dc opteron running FC5 and amd64 3000+ running rawhide)
 >  > the last one (-m64) is weird (much slower!!) no -OX was used.

When run in 32bit, this is doubling the amount of loads/stores used,
so little wonder it's slower.
what do you mean by "when run in 32bit" ? the resulting binary was a x86_64 one ??
		Dave



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]